基于抑制性突触多层神经元群放电编码的图像边缘检测

doi:10.3969/j.issn.0258-8021.2014. 05.01

摘要
图/表
参考文献
相关文章 (2)

全文: PDF (2871 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要图像边缘检测所获得的目标轮廓细节质量，对于后续图像分析或理解过程具有重要作用。提出一种基于视觉机制的图像边缘检测新方法:构建抑制性突触连接的多层神经元群，在待检测图像的激励下，分析7像素×7像素视觉感受野窗口内互连接神经元的脉冲放电过程，记录发放时刻进行次序编码；同时考虑神经元之间的侧向抑制作用，在选择注意机制作用下获得增强图像；之后利用LogGabor滤波器模拟视觉系统中的方向选择特性，获取8个方向的滤波结果，经过输出层神经元群的融合处理并经灰度映射到0～255的范围后获得边缘图像。对含有丰富边缘细节特性的24幅菌落图像进行处理，其处理结果的ROC评价指标均值为0.698 4，优于PCNN法的0.659 3；从评价指标的均方差来看，该结果具有更好的一致性。另外，从信息熵评价指标来看，该方法同样具有一定的优势，能够有效提取图像的边缘信息，而且也能反映更多层次的图像细节。所提出的方法为利用视觉生理特性进行图像处理提供了崭新而有效的思路。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	廖进文范影乐^*武薇高云园李轶

关键词 ：边缘检测, 神经元群, 抑制性突触, 选择注意机制, ROC指标

Abstract：Details of objects’ contour can be acquired from the edge detection, which are considered important for image analysis and understanding correctly. In this paper, a new method for image edge detection was proposed based on vision mechanism. Multilayer neuronal population with inhibitory synapse was constructed to receive the stimuli from an awaited image, and the process of pulse spiking from connected neurons in the 7×7 window of visual receptive field was analyzed, spiking times were recorded for rank coding.Considering the effect of lateral inhibition between neurons, images with the mechanism of selective attention were enhanced; then LogGabor filter was adopted to stimulate the orientation selectivity of visual system for obtaining the filtering results of 8 orientations, the edge map could be acquired through integration processing of the output neuron layer and gray mapping to the range between 0 and 255. Taken the colony images with abundant edge details for processing, the mean value of ROC index for the results was 0698 4, better than that of PCNN, which was 0659 3; the results of the new method were better in the view of consistency in terms of the mean square deviation. Additionally, the new method also owning certain advantages in terms of information entropy, indicating the method proposed could extract edge information effectively and reflect image details in more levels. The method of edge detection proposed in the paper provides a new and effective idea for the image processing based on visual physiological characteristics.

Key words： edge detection neuronal population inhibitory synapse selective attention mechanism ROC index

基金资助:国家自然科学基金(61201300)；浙江省教育厅科研项目（Y201328513）；浙江省人事厅留学回国人员科研启动基金

引用本文:

廖进文范影乐^*武薇高云园李轶. 基于抑制性突触多层神经元群放电编码的图像边缘检测[J]. 中国生物医学工程学报, 2014, 33(5): 513-524.
LIAO Jin Wen FAN Ying Le^*WU Wei Gao Yun Yuan LI Yi.

Image Edge Detection Based on Spike Coding of Multilayer Neuronal Population with Inhibitory Synapse

. journal1, 2014, 33(5): 513-524.

链接本文:

http://cjbme.csbme.org/CN/10.3969/j.issn.0258-8021.2014. 05.01 或 http://cjbme.csbme.org/CN/Y2014/V33/I5/513

［1］Maini R, Aggarwal H. Study and comparison of various image edge detection techniques［J］. International Journal of Image Processing (IJIP), 2009, 3(1): 1-11.
［2］Juneja M, Sandhu PS. Performance evaluation of edge detection techniques for images in spatial domain［J］. Methodology, 2009, 1(5): 614-621.
［3］Kerr D, Coleman S, McGinnity M, et al. Biologically inspired edge detection［C］// 2011 11th International Conference on Intelligent Systems Design and Applications (ISDA). Cordoba：IEEE, 2011: 802-807.
［4］Wu Qingxiang, McGinnity M, Maguire L, et al. Edge detection based on spiking neural network model［M］//Advanced Intelligent Computing Theories and Applications. With Aspects of Artificial Intelligence. Berlin：Springer Berlin Heidelberg, 2007: 26-34.
［5］Meftah B, Lezoray O, Benyettou A. Segmentation and edge detection based on spiking neural network model［J］. Neural Processing Letters, 2010, 32(2): 131-146.
［6］Ba瘙塂türk A, Günay E. Efficient edge detection in digital images using a cellular neural network optimized by differential evolution algorithm［J］. Expert Systems with Applications, 2009, 36(2): 2645-2650.
［7］Zhou Liang, Sun Yu, Zheng Jianguo. Automated color image edge detection using improved PCNN model［J］. WSEAS Transactions on Computers, 2008, 7(4): 184-189.
［8］Coombes S, Thul R, Wedgwood KCA. Nonsmooth dynamics in spiking neuron models［J］. Physica D: Nonlinear Phenomena, 2012, 241(22): 2042-2057.
［9］Izhikevich EM. Which model to use for cortical spiking neurons?［J］. IEEE Transactions on Neural Networks, 2004, 15(5): 1063-1070.
［10］Izhikevich EM. Simple model of spiking neurons［J］. IEEE Transactions on Neural Networks, 2003, 14(6): 1569-1572.
［11］Clopath C, Büsing L, Vasilaki E, et al. Connectivity reflects coding: a model of voltagebased STDP with homeostasis［J］. Nature Neuroscience, 2010, 13(3): 344-352.
［12］Nessler B, Pfeiffer M, Maass W. STDP enables spiking neurons to detect hidden causes of their inputs［C］//Advances in Neural Information Processing Systems. Vancouver: NIPS, 2009: 1357-1365.
［13］Jonas P, Buzsaki G. Neural inhibition［J］. Scholarpedia, 2007, 2(9): 3286. 
［14］Bruce NDB, Tsotsos JK. Saliency, attention, and visual search: An information theoretic approach［J］. Journal of Vision, 2009, 9(3).
［15］Itti L, Koch C. Computational modelling of visual attention［J］. Nature Reviews Neuroscience, 2001, 2(3): 194-203.
［16］Fix J, Rougier N, Alexandre F. A dynamic neural field approach to the covert and overt deployment of spatial attention［J］. Cognitive Computation, 2011, 3(1): 279-293.
［17］Itti L, Koch C, Niebur E. A model of saliencybased visual attention for rapid scene analysis［J］.
IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998,20(11): 1254-1259.
［18］RamirezMoreno DF, Schwartz O, RamirezVillegas JF. A saliencybased bottomup visual attention model for dynamic scenes analysis［J］. Biological Cybernetics, 2013, 107: 141-160.
［19］AvilaAkerberg O, Chacron MJ. Nonrenewal spike train statistics: causes and functional consequences on neural coding［J］. Experimental Brain Research, 2011, 210(3-4): 353-371.
［20］Thorpe SJ, Gautrais J. Rapid visual processing using spike asynchrony［J］. Advances in Neural Information Processing Systems, 1997: 901-907.
［21］Thorpe SJ. Spikebased image processing: can we reproduce biological vision in hardware?［C］//Proceedings of the 12th International Canferece on Computer Vision. Berlin :Springer Berlin Heidelberg, 2012: 516-521.
［22］Isaacson JS, Scanziani M. How inhibition shapes cortical activity［J］. Neuron, 2011, 72(2): 231-243.
［23］Alitto HJ, Dan Y. Function of inhibition in visual cortical processing［J］. Current Opinion in Neurobiology, 2010, 20(3): 340-346.
［24］Sadagopan S, Ferster D. Feedforward origins of response variability underlying contrast invariant orientation tuning in cat visual cortex［J］. Neuron, 2012, 74(5): 911-923.
［25］Bowyer K, Kranenburg C, Dougherty S. Edge detector evaluation using empirical ROC curves［J］. Computer Vision and Image Understanding, 2001, 84(1): 77-103.
［26］Yitzhaky Y, Peli E. A method for objective edge detection evaluation and detector parameter selection［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(8): 1027-1033.
［27］FernándezGarcía NL, CarmonaPoyato A, MedinaCarnicer R, et al. Automatic generation of consensus ground truth for the comparison of edge detection techniques［J］. Image and Vision Computing, 2008, 26(4): 496-511.
［28］Fawcett T. An introduction to ROC analysis［J］. Pattern Recognition Letters, 2006, 27(8): 861-874.
［29］Brink AD. Using spatial information as an aid to maximum entropy image threshold selection［J］. Pattern Recognition Letters, 1996, 17(1): 29-36.
［30］Verma OP, Hanmandlu M, Kumar P, et al. A novel bacterial foraging technique for edge detection［J］. Pattern recognition letters, 2011, 32(8): 1187-1196.